Download Substrate Specificity and Mechanism from the Structure of

doi:10.1016/j.jmb.2004.01.043
J. Mol. Biol. (2004) 337, 387–398
Substrate Specificity and Mechanism from the
Structure of Pyrococcus furiosus Galactokinase
Andrew Hartley1, Steven E. Glynn1, Vladimir Barynin1
Patrick J. Baker1, Svetlana E. Sedelnikova1, Corné Verhees2
Daniel de Geus2, John van der Oost2, David J. Timson3
Richard J. Reece4 and David W. Rice1*
1
Krebs Institute, Department
of Molecular Biology and
Biotechnology, University
of Sheffield, Sheffield S10 2TN
UK
2
Laboratory of Microbiology
Wageningen University
Hesselink van Suchtelenweg 4
NL-6703 CT Wageningen, The
Netherlands
3
Medical Biology Centre
School of Biology and
Biotechnology, Queen’s
University Belfast, 97 Lisburn
Road, Belfast BT9 7BL, UK
4
School of Biological Sciences
University of Manchester
2.205 Stopford Building
Oxford Road, Manchester M13
9PT, UK
*Corresponding author
Galactokinase (GalK) catalyses the first step of the Leloir pathway of
galactose metabolism, the ATP-dependent phosphorylation of galactose
to galactose-1-phosphate. In man, defects in galactose metabolism can
result in disorders with severe clinical consequences, and deficiencies in
galactokinase have been linked with the development of cataracts within
the first few months of life. The crystal structure of GalK from Pyrococcus
furiosus in complex with MgADP and galactose has been determined to
2.9 Å resolution to provide insights into the substrate specificity and catalytic mechanism of the enzyme. The structure consists of two domains
with the active site in a cleft at the domain interface. Inspection of the substrate binding pocket identifies the amino acid residues involved in galactose and nucleotide binding and points to both structural and mechanistic
similarities with other enzymes of the GHMP kinase superfamily to which
GalK belongs. Comparison of the sequence of the Gal3p inducer protein,
which is related to GalK and which forms part of the transcriptional
activation of the GAL gene cluster in the yeast Saccharomyces cerevisiae,
has led to an understanding of the molecular basis of galactose and
nucleotide recognition. Finally, the structure has enabled us to further
our understanding on the functional consequences of mutations in
human GalK which cause galactosemia.
q 2004 Elsevier Ltd. All rights reserved.
Keywords: galactokinase deficiency; GHMP kinase superfamily; Pyrococcus
furiosus; X-ray crystallography
Introduction
Pyrococcus furiosus is a hyperthermophilic
archaeon, isolated from thermally active regions.
Like other saccharolytic organisms, P. furiosus
Abbreviations used: FA, structure factor amplitudes for
the anomalous scatterers; GalK, galactokinase; GHMP,
galactokinase homoserine kinase, mevalonate kinase
phosphomevalonate kinase; GTP, guanosine-50 triphosphate; HSK, homoserine kinase; MAD, multiwavelength anomalous dispersion; MMK, Methanococcus
janaschii mevalonate kinase; NCS, non-crystallographic
symmetry; Pf, Pyrococcus furiosus; RMK, Rattus norvegicus
mevalonate kinase; rmsd, root-mean-squared deviation;
SeMet, selenomethionine; s, standard deviation.
E-mail address of the corresponding author:
[email protected]
utilises the glycolytic pathway as a main catabolic
route. Conversion of galactose via glycolysis
requires an additional metabolic branch, the Leloir
pathway,1 and the first step of this pathway
involves the conversion of galactose to galactose1-phosphate, catalysed by galactokinase (GalK; EC
2.7.1.6), which drives the transfer of the g-phosphate group from ATP to the O1 position of
galactose.
In man, galactosemia is an autosomal recessive
disorder caused by a defect in one of the three
Leloir pathway enzymes involved in galactose
metabolism, galactokinase, galactose-1-phosphate
uridyl transferase (GalT) and UDP-galactose 40
epimerase (GalE).2 Deficiency in the transferase,
GalT, is responsible for classic galactosemia (MIM
230400) and can lead to severe neonatal symptoms
0022-2836/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.
388
Structure of P. furiosus Galactokinase
Figure 1. Structure-based sequence alignment of representative members of the GHMP family, galactokinase family
and yeast Gal3p. GalK, galactokinase; HSK, Methanococcus janaschii homoserine kinase; RMK, Rattus norvegicus mevalonate kinase; MMK, M. janaschii mevalonate kinase; Pf, Pyrococcus furiosus; Hs, Homo sapiens; Ec, Escherichia coli; Sp,
389
Structure of P. furiosus Galactokinase
such as failure to thrive, hepatomegaly and
bacterial sepsis. Galactokinase deficiency (galactosemia II; MIM 230200) in man is an inborn error of
galactose metabolism and is linked to development
of cataracts during the first months of life and also
pre-senile cataracts, the onset of which is between
20 and 50 years of age.3 Over 20 mutations have
been identified in human galactokinase that are
associated with reduced galactokinase activity in
the blood,4 – 8 11 of these mutations are single
amino acid substitutions and the biochemical
characteristics of some of these have recently been
characterised.9
Studies on the yeast Saccharomyces cerevisiae have
led to identification of a galactokinase (Gal1p) in
this organism as part of a gene cluster which
encodes the enzymes of the Leloir pathway.10,11
The transcriptional activation of this cluster, which
results in Gal1p expression involves three proteins:
a transcriptional activator and DNA-binding
protein, Gal4p; a repressor, Gal80p; and a ligand
sensor and inducer, Gal3p. Induction occurs as a
result of an ATP and galactose-dependent interaction between Gal3p and Gal80p with Gal3p
acting as the ligand sensor and transducer of galactose signal.12 Sequence analysis has shown that
there is a very high similarity (73% identity)
between the yeast proteins Gal1p and Gal3p.13
However, despite this similarity, Gal3p does not
show galactokinase activity and the transcriptional
activation of the gene cluster by Gal1p is also
observed, although at a much reduced level compared to Gal3p.13 Determination of the structure of
Pyrococcus furiosus (Pf) GalK may offer insights
into the mechanism of galactose and ATP binding
in the yeast Gal3p protein and also shed light on
the interactions involved in the Gal3p – Gal80p
complex.
Sequence similarities have suggested that GalK
is a member of the GHMP kinase superfamily14
for which the structures of several family members
have now been reported, including homoserine
kinase from Methanococcus janaschii15 (HSK; EC
2.7.1.39),
mevalonate
kinase
from
Rattus
norvegicus16 (RMK; EC 2.7.1.36), and from
M. janaschii17 (MMK), mevalonate diphosphate
decarboxylase from M. janaschii18(MDD; EC
2.7.4.2) and phosphomevalonate kinase from
Streptococcus pneumoniae19 (PMK; EC 2.7.4.2).
Although the overall sequence identity in the
family is low20 (with 16%, 15% and 20% identities
between Pf GalK and HSK, RMK and MMK,
respectively), three conserved glycine-rich motifs
can be identified (motifs 1 – 3; Figure 1). The threedimensional structure of homoserine kinase was
the first to be solved for any member of this family
and reveals a novel nucleotide binding fold and a
less common, syn conformation of the glycosidic
bond of the ADP in the active site.20 However,
determination of the structure of another GHMP
kinase family member, mevalonate kinase,16 in
complex with ATP, revealed that in some enzymes
of the family, the glycosidic bond can adopt an
anti conformation. Analysis of the structures of the
GHMP kinase superfamily has shown that the
most conserved motif across the GHMP kinase
family, with a consensus of PX3GL(G/S)SSA
(motif 2; Figure 1), forms an atypical phosphate
binding P-loop which interacts through hydrogen
bonds to the a and b-phosphate groups of ADP.
More recently, the three-dimensional structure of
the GalK from Lactobacillus lactis in complex with
galactose and inorganic phosphate has been
reported. This structure confirms the general similarity of the GalK fold to the other members of the
GHMP kinase superfamily.21
Here, we report the three-dimensional structure
of Pf GalK at 2.9 Å resolution in a complex with
ADP and galactose. The structure reveals the
mode of nucleotide recognition and its relationship
to the galactose-binding site. Comparison of the
structure with other members of the GHMP kinase
superfamily has permitted the identification of the
residues involved in substrate binding and
catalysis and led to a better understanding of the
lack of galactokinase activity in Gal3p. The
sequence similarity between Pf GalK and human
GalK has facilitated a homology-based modelling
study, which has shed light on the structural basis
of the galactosemia causing amino acid substitutions in the human GalK protein.
Results and Discussion
Quality of the structure
The GalK structure was solved using multiwavelength anomalous dispersion (MAD) data
collected on a single selenomethionine labelled
(SeMet) crystal to 2.9 Å resolution. The final
refined model of GalK contains nine independent
subunits in the asymmetric unit (labelled A
through I in Figure 2(a)), a total of 2999 amino
acid residues. Analysis of the packing in the cell
reveals subunit I is related by a crystallographic
2-fold axis to its symmetry related partner, I0 and
subunits A and D appear to be related by an
approximate non-crystallographic 2-fold axis of
symmetry within the crystal. However, the
Streptococcus pneumoniae; Sc, Saccharomyces cerevisiae. The invariant residues are highlighted in black with identities in
7/9 sequences highlighted in dark grey. Identities in 3/4 of the non-yeast galactokinases are highlighted in light grey.
The secondary structural elements in the Pf GalK are shown above the sequences. Invariant residues amongst the
galactokinase family are boxed. The mutations identified in galactokinase deficiency in man are identified above the
sequences with the corresponding substituted amino acid single letter code.
390
Structure of P. furiosus Galactokinase
Figure 2. (a) Schematic diagram of the crystal packing arrangement of GalK. The positions of all nine monomers are
shown and labelled A to I. (b) Schematic diagram of the GalK monomer. The N-terminal domain is shown in blue and
the C-terminal domain in green. The three conserved motifs in the GHMP kinase superfamily are highlighted (1, red; 2,
beige; 3, gold). Helices are labelled aA through aJ and b-strands labelled B1 to B13. ADP and galactose are represented
with a stick diagram (atom-coloured). Mg2þ is coloured cyan. Both the N and C termini are labelled. Label colours are
not significant.
interfaces that mediate the interactions between
these pairs of subunits are different and the fraction of the solvent accessible surface buried across
both interfaces (3% and 4.5% for subunits I– I0 and
A – D, respectively) lies below the minimum range
found in an analysis of other dimeric proteins.22
Gel filtration of Pf GalK resulted in elution at the
approximate molecular mass of the monomer
(40 – 45 kDa compared to the calculated molecular
mass of 39.4 kDa, SWISSPROT: Q9HHB6,23,24),
suggesting that the monomer described here is the
biologically relevant oligomerization state. The
path of the polypeptide chain is well defined by
the electron density map for five subunits.
Structure of P. furiosus Galactokinase
However, in two of the other subunits 4% and 2%
of residues have very poor density for the mainchain and in the other two subunits, 12% and 27%
of the residues corresponding to part of the
N-terminal domain have very poor density for the
main-chain. Superposition of the Ca atoms of
the subunits using the program LSQKAB25 gives
an average rmsd value for the Ca atoms of 0.23 Å
showing that there are no major conformational
differences in the subunits.
Overall polypeptide fold
The following discussion is based on an analysis
of subunit I, the best-defined subunit in the
electron density. The GalK monomer is approximately 60 Å £ 55 Å £ 40 Å in size, comprising two
domains separated by a deep cleft. The N-terminal
domain (Figure 2(b)) is composed of a mixed, fivestranded b-sheet (b1, b4– b7) packed on one face
by a bundle of four helices (aA – aD) with the
other face exposed entirely to solvent. A small antiparallel b-sheet (b2, b3, b8, b9) that lies at the
C-terminal end of a five-stranded b-sheet dominates the domain interface. The C-terminal domain
(Figure 2(b)) is composed of an antiparallel b-sheet
(b10– b13) and five a-helices (aE –aH) with the
b-sheet completing the domain interface.
Nucleotide binding site
Crystallisation of Pf GalK was achieved in the
391
presence of ADP (5 mM), MgCl2 (10 mM) and
galactose (20 mM).23 A difference Fourier map calculated at a late stage of the refinement revealed
two areas of additional electron density in the cleft
between the two domains. This density lies in a
similar position to the nucleotide and substrate
binding sites of the homologous GHMP kinase
family members and an ADP and galactose molecule were fitted into this density and refined. The
quality of the electron density compared to the surrounding protein is such that we estimate the
ADP/Mg2þ and galactose sites are fully occupied
(Figure 3(a) and (b)). The ADP molecule is bound
in an anti conformation with the adenine ring
situated in a hydrophobic pocket formed by the
side-chains of residues Phe52, Trp69, Ile94, Phe110
and Leu100 (Figure 4(a)). This anti conformation
of the adenine ring contrasts with the syn conformation proposed in the recent structure determination of GalK from L. lactis on the basis of
modelling.21 The adenine ring of ADP forms a p
face-to-face stacking interaction with the aromatic
ring of Phe110 on one side with the other face
packing against the side-chain of Leu100
(Figure 4(a)). The N1 nitrogen atom on the adenine
ring makes a hydrogen bond with the hydroxyl
oxygen atom of Ser49. There are no other direct
hydrogen bonds between protein atoms and the
nitrogen atoms on the adenine ring. However,
there is space in the structure for water molecules
to be involved in water-mediated hydrogen bonds
between the N6 amino group and main-chain
Figure 3. (a) Final electron density map ð2Fo 2 Fc Þ in the vicinity of the galactose-binding site contoured at the 1s
level. The galactose molecule and surrounding protein residues are labelled and shown in stick format. (b) Final
electron density map, as in (a), in the vicinity of the enzyme-bound MgADP. The ADP molecule is shown as sticks
and the magnesium ion is shown with an orange ball.
392
Structure of P. furiosus Galactokinase
Figure 4. (a) Stereo diagram of the ADP-binding site showing the ADP and galactose (green and atom colours) and
the protein backbone of nearby residues in blue. Residues involved in direct interactions to the ADP are shown in
stick format (grey and atom colours). Hydrogen bonds to the ADP and ion pair interactions to the Mg2þ are shown
as blue broken lines. (b) Stereo diagram of the galactose-binding site showing galactose (grey and atom colours),
Mg2þ and part of the ADP (green and atom colours). The nearby residues involved in direct interactions with the
galactose and hydrogen bonds to the galactose are shown as for (a).
atoms. Biochemical studies have shown that Pf
GalK will also utilise GTP as a phosphoryl donor
with only a 40% decrease in activity compared to
ATP.24 This is consistent with the analysis of the
purine-binding pocket, which suggests that a GTP
molecule could be accommodated with only
minor modifications to the position of the guanine
ring. The ribose moiety packs against the sidechain aromatic ring of Tyr72 with its 20 and 30
hydroxyl groups exposed to the solvent and not
forming any direct hydrogen bonds to the enzyme.
The residues 99PLGAGLSSSAS109 corresponding to
motif 2 (Figure 1), form a phosphate-binding loop
which wraps around the b-phosphate group, forming hydrogen bonds with the main-chain amide
groups of Gly101 and Ser107 and the side-chain
hydroxyl groups of Ser106 and Ser107. At this resolution, we cannot unambiguously identify the
position of the bound magnesium, which we
assume is part of the complex given its inclusion
in the crystallisation mixture. However, the
electron density shows a bulge between the a and
b-phosphate groups (Figure 3(b)), the position of
which maps exactly on to that identified as the
divalent cation in the structures of the other
GHMP family members. The magnesium in the
GalK structure is coordinated by the side-chain
oxygen atoms of the conserved residues, Glu130
and Ser107 and by oxygen atoms from the a and
b-phosphate groups of ADP (Figure 4(a)).
Galactose binding
The galactose sits in a cavity between the two
domains, oriented with the phosphoryl acceptor
oxygen atom (O1) towards the phosphate-binding
loop and the position of the terminal phosphate of
the bound ADP nucleotide (Figure 4(b)). The cavity
is formed from and surrounded by residues conserved amongst galactokinases (motifs 1 and 3; see
Figure 1) and these form extensive and specific
interactions with the galactose oxygen atoms. The
O6 hydroxyl of galactose forms hydrogen bonds
with a carboxyl oxygen atom of Glu17 and the
main-chain amide of His18. The O4 atom of galactose forms a hydrogen bond with the hydroxyl
group of Tyr200. The carboxyl oxygen atoms of
Asp20 hydrogen bond with both the O4 and O3
hydroxyl oxygen atoms. The carboxyl oxygen of
Asp151 is in a position to form a hydrogen bond
with the O2 hydroxyl of galactose and is 3.4 Å
from the O1 hydroxyl group, which is phosphorylated by GalK. Arg11 also forms a hydrogen bond
with this phosphoryl acceptor hydroxyl. The
Structure of P. furiosus Galactokinase
interactions made by galactose in Pf GalK are
similar to those made by galactose in L. lactis
GalK.21
Comparison with other GHMP kinase
family members
The overall structure of GalK is very similar to
the structures of the other GHMP kinase superfamily that have been solved to date, even though
overall sequence similarity within the family is
low. The conserved motifs, 1– 3 (Figure 1), show
structural similarity to the other GHMP family
members with motifs 1 and 3 forming part of the
galactose-binding site and motif 2 forming the
phosphate-binding loop. However, there are some
differences in structure between GalK and the
other GHMP kinase family members. The anti conformation of the ADP bound to GalK is similar to
that observed in the complex of RMK and ATP16
but is in contrast to the syn conformation of the
nucleotide found in the equivalent complex with
HSK.15 However, the binding of the adenine ring
in GalK is different to the RMK structure as there
are no direct hydrogen bonds involving the amino
group of the adenine ring in the GalK structure.
Instead, it is possible that water-mediated hydrogen bonds are involved. The region corresponding
to helices aE and aF in Pf GalK (residues
195– 231) does not show similarity to any of the
other GHMP kinase structures and a DALI search
reveals no other structures with a Z-score of over
2.5, suggesting that this may be a novel structural
motif.
GalK catalytic mechanism
Previous studies on GHMP kinase superfamily
members have led to two distinct proposals for
the catalytic mechanism. Phosphoryl transfer in
393
HSK has been suggested to occur by direct nucleophilic attack on the g-phosphate group of ATP by
the d-hydroxyl of homoserine.20 In this mechanism,
the latter is stabilised by the formation of a hydrogen bond to a neighbouring asparagine residue
(Asn141), not conserved in the superfamily. Catalysis is proposed to be assisted through activation of
the g-phosphate of ATP by the magnesium ion,
which is coordinated by a conserved glutamate
residue (Glu130) with the deprotonation of the
d-hydroxyl possibly involving the g-phosphate.20
In contrast, studies on both RMK16 and MMK17
have suggested a mechanism involving an aspartic
acid residue conserved in these two enzymes as a
possible catalytic base (Asp204 in RMK) whose
role is to activate the receptor hydroxyl group. A
conserved lysine residue (Lys13 in RMK) is then
proposed to stabilise the resultant C5 alkoxide
group and the penta-coordinated g-phosphate.
Comparison of the structures of HSK, RMK and
GalK reveals the presence of Arg11 and Asp151 of
GalK in positions equivalent to the proposed
catalytic lysine and aspartate residues, respectively,
in the mevalonate kinases (Figure 5). Asp151 is
located 3.8 Å from the O1 hydroxyl of galactose,
suggesting that this residue is well placed to play
a possible role in proton abstraction. Arg11 is
located 3.2 Å from the O1 phosphoryl acceptor
hydroxyl with its guanidinyl group suitably positioned to stabilise the developing negative charge
which accumulates during the transition state. As
observed in RMK, MMK and HSK, a glutamate
residue is involved in coordination of the magnesium ion (Glu193 in RMK, Glu130 in HSK and
Glu139 in GalK). Studies on HSK have suggested
that Gln152 in GalK may have a comparable role
to Asn141 in HSK, responsible for stabilisation of
the acceptor hydroxyl group.20 However, superposition of the structures shows that Asn141 in
HSK is equivalent to the putative catalytic
Figure 5. Stereo diagram showing a superposition of the active site residues of GalK (grey), HSK (red) and RMK
(green). The galactose and ADP molecules as well as the magnesium ion from the GalK structure are also shown.
The superposition shows the equivalence in position of the proposed catalytic residues in GalK and RMK (Asp151/
Asp204 and Arg11/Lys13, respectively). Notable differences in these residues seen in HSK, suggest that there may be
differences in the catalytic mechanism of the latter. The disparity in the positions of the proposed catalytic asparagine
in HSK (Asn141) and Gln152 of GalK is clearly shown.
394
aspartate residues in GalK and RMK (Asp151 and
Asp204, respectively) (Figure 5). Moreover, the
amide group of Gln152, being the next residue in
this helical section of the molecule, is some 9 Å
away from the position of the O1 hydroxyl of
galactose and not in a suitable orientation to be
considered a potential catalytic residue. In
addition, HSK is the only enzyme in the family
that does not possess a positively charged residue
equivalent to Lys13 (RMK) and Arg11 (GalK) and
instead, HSK has a threonine (Thr14) at this
position (Figure 5). This would suggest that the
mechanistic details of GalK are more closely
related to the mevalonate kinases and possibly
distinct to HSK.
Comparison with S. cerevisiae GalK and Gal3p
The structure-based sequence alignment of the
galactokinase family reveals significant sequence
similarity between Pf GalK and the yeast
S. cerevisiae galactokinase (Gal1p) and Gal3p
inducer protein (Figure 1). For Gal1p and Gal3p,
the location of the strongly conserved residues is
virtually identical to that seen across the galactokinase family. Thus, of the seven residues
intimately involved in galactose binding in the Pf
enzyme, six can be identified in the Gal1p and
Gal3p proteins, the only exception being the
replacement of Tyr200 by a lysine in both. Similarly, of the five residues involved in ATP binding
and conserved across the galactokinase family,
four can be identified in Gal3p. This implies that
the recognition of galactose and ATP by these proteins will be closely related. Compared to the Pf
enzyme both of the yeast proteins are considerably
larger with the difference being primarily due to an
Structure of P. furiosus Galactokinase
insertion of 93 residues that is located in the loop
between helices aE and aF where it would not
interfere with the basic fold of the protein.
Wild-type Gal3p lacks galactokinase activity
despite sharing most of the substrate binding
motifs with GalK enzymes. However, Gal3p does
have a two-residue deletion within motif 3
(Figure 1) compared to other members of the
family, corresponding to the loss of Ser107 and
Ala108 in the Pf structure. In Pf GalK Ser107 interacts with both the magnesium ion and b-phosphate of ATP. Gal3p is only functionally active in
the presence of a nucleotide (ATP or ADP) and
magnesium26,27 and therefore it is likely that this
deletion affects the relative positioning of the
g-phosphate group of the ATP with respect to the
O1 hydroxyl of galactose rather than preventing
substrate binding. This is consistent with the biochemical data showing that the insertion of serine
and alanine residues at this position in Gal3p confers galactokinase activity, although at a reduced
level when compared to Gal1p13.
Insights into galactosemia inducing mutations
of GalK
The structure of Pf GalK, together with the close
sequence similarity amongst the galactokinase
family has allowed us to investigate the location
and effect of the mutations in the monomeric
human galactokinase9 that cause galactosemia,
comparing earlier conclusions drawn from the
structure of L. lactis GalK.21
P28T is a mutation causing galactosemia found
amongst Roma gypsies,5 and corresponds to Ser2
in the Pf structure (Figure 6), which is at the start
of strand b1 in the N terminus. The human GalK
Figure 6. Schematic stereo representation of the Ca backbone of the GalK monomer with invariant residues in the
galactokinase family coloured red. The positions of the amino acid substitutions in human GalK responsible for
galactosemia are mapped onto the Pf GalK and shown as yellow spheres and the corresponding labels indicate the
mutations in human GalK.
Structure of P. furiosus Galactokinase
has a 25 amino acid insertion before the Pf N
terminus so it is likely that this area differs in
structure, preventing a detailed comparison.
V32M, another common mutation causing
galactosemia,7 corresponds to Val6 in the Pf structure. Val6 is located on strand b1 which makes up
the N-terminal b-sheet that is conserved amongst
members of the GHMP family, and the side-chain
of this residue is buried in a hydrophobic cluster
which involves residues from helix aC. This helix
is important in providing contacts to the magnesium ion through the side-chain of Ser107 and
also provides a dipole charge stabilisation of the
negative charge of the pyrophosphate moiety of
the nucleotide. The substitution of valine by a
bulkier methionine residue in this buried part of
the molecule is likely to cause substantial disruption to the structure.
G36R corresponds to Gly10 in Pf GalK, a highly
conserved residue in motif 1 (Figure 1), which
packs against the phosphate-binding loop. Substitution to an arginine residue would disturb this
packing interaction and cause alterations in the
structure of the nucleotide-binding site. This
mutation, along with P28T and V32M, when made
in human GalK caused the protein to be insoluble
upon expression in Escherichia coli, suggesting that
misfolding or destabilisation of the folded state is
caused by these mutations.9
H44Y corresponds to a conserved histidine residue in Pf GalK, His18, which is located in motif 1
which forms a loop between b2 and a 310 turn
forming part of the galactose binding pocket. The
N11 and N12 nitrogen atoms of this histidine form
hydrogen bonds to the main-chain carbonyl
oxygen atom of Ala305 and the side-chain
hydroxyl oxygen of Tyr21, respectively, and the
main-chain carbonyl oxygen forms a hydrogen
bond with the O6 atom of galactose. Substitution
of this buried histidine residue to a bulkier tyrosine
would result in loss of these interactions but
probably more importantly, the pocket that the histidine side-chain sits in cannot accommodate the
larger tyrosine ring. This would cause disruption
of the galactose-binding pocket, consistent with
the reduced Km value for galactose and lower kcat
value found in the human enzyme with this
mutation.9
R68C corresponds to Tyr42 in Pf GalK, which is
situated on a large loop between b3 and b4 on the
solvent exposed surface of the N-terminal domain.
This residue, however, is located over 30 Å from
the active site and whilst it is conserved between
the human, E. coli and S. pneumoniae galactokinases, it is surprising that a substitution so
remote can have an impact on the functionality of
the enzyme.9
A198V corresponds to Val163 in Pf GalK and is
located on strand b8, which forms part of the antiparallel b-sheet in the N-terminal domain. Human
GalK with the A198V mutation has kinetic properties that are almost identical to wild-type human
GalK.9 This is consistent with the observation that
395
the site of this mutation is distant from the active
site and the conservative nature of the mutation in
a relatively uncrowded environment would
probably have little effect on the function of this
enzyme. This is further supported by an analysis
of the levels of erythrocytic galactokinase in
patients with this mutation, which parallels the
reduction in kcat8 value, suggesting that this
mutation does not affect the function of
galactokinase.
T288M corresponds to Val248 in Pf GalK and is
located in the middle of aG with its side-chain
packed against the C-terminal b-sheet and particularly close to residue Gly158, which is conserved
amongst the galactokinase family. Gly158 immediately follows helix aD that forms one part of the
active site and contains the potential catalytic
residue, Asp151. Creation of the T288M mutation
in human GalK caused the protein to be
insoluble on expression,9 suggesting that perturbation of this packing interaction causes
misfolding of GalK.
G346S appears in a loop between strands b11
and b12, which is an area of high sequence similarity across the GHMP kinase family where it
forms part of motif 3, corresponding to Ala305 in
Pf GalK. This residue is either an Ala or Gly in the
aligned GHMP family members and galactokinase
family (Figure 1) and in the Pf structure the mainchain carbonyl group of Ala305 forms a hydrogen
bond to His18, which itself forms an important
interaction with the galactose. The side-chain of
Ala305 points towards the solvent but with the
b-carbon atom only 4.7 Å from the O1 oxygen of
galactose. Whilst it appears from the analysis of
the structure that the substitution to serine can be
accommodated with the additional hydroxyl not
forming any adverse steric interactions, the likely
structural changes that accompany the attack of
the O1 oxygen atom on the g-phosphate of ATP
may well be affected. This is consistent with the
finding that this mutant showed little change in
the Km value for galactose but the kcat =Km;gal value
was reduced 100-fold compared to the wild-type
GalK.9
G349S corresponds to Gly308 in Pf GalK, which
is a completely conserved residue in motif 3
lying close to the active site. This residue lies in a
region of the Ramachandran plot that is unfavourable for residues other than glycine. The
observed reduction in kcat resulting from this
mutation9 may arise as a result of the enforced
differences in the Ramachandran angles adopted
in the mutant structure and a possible reduction
in flexibility, affecting important transition state
interactions.
A384P corresponds to S347 in Pf GalK and is
located in the C-terminal tail away from the active
site. However, this is an area of low sequence
similarity in the galactokinase family and therefore
it is not possible to accurately describe the effect
that this mutation may have in the human
homologue.
396
Conclusion
The structure of Pf GalK presented here represents the first for the ternary complex of this
enzyme and the first description of any GalK from
a thermophilic organism. The structure has
revealed the molecular basis for the recognition of
galactose, ADP and Mg2þ and has provided
insights in to the enzyme mechanism. Analysis of
the structure and comparison with other GHMP
kinase family members has revealed considerable
similarities in nucleotide binding and catalysis. In
particular the mode of nucleotide binding is
closely related to that seen in RMK and MMK
with an anti conformation of the glycosidic bond
unlike the syn conformation observed in HSK. In
mechanistic terms, the identification of conserved
negatively and positively charged residues
(Asp151 and Arg11, respectively) in equivalent
positions to those proposed to act as a catalytic
base and to stabilise the negative charges generated at the catalytic centre in the mevalonate
kinases (Asp204 and Lys13 in RMK), points to
clear similarities in mechanism. Finally, the
structure of Pf GalK has been useful in extending
our ideas on the interactions which control ATP
and galactose recognition in the yeast protein
Gal3p and the consequences of mutations which
lead to galactosemia in man.
Materials and Methods
SeMet incorporation, over-expression, purification and
crystallisation of Pf GalK were performed as described.23
For data collection, described in detail by De Geus et al.,23
the crystal was transferred to a cryoprotectant solution
containing the original precipitant solution plus 20%
(v/v) glycerol and flash frozen at 100 K. A single crystal
was used for MAD experiments performed at three
wavelengths on beamline ID14.4 at the ESRF Grenoble
laboratory at 100 K. The data for each wavelength were
processed and merged using the DENZO/SCALEPACK
package.28 The crystals belong to space group C2221
with cell dimensions a ¼ 211:7 Å, b ¼ 355:4 Å,
c ¼ 165:5 Å, a ¼ b ¼ g ¼ 908.
The diffraction data were analysed and FA values
determined using the program XPREP (Bruker AXS,
Madison, USA). The anomalous signal was found to
extend to 2.9 Å and FA values were truncated to this
resolution. The SeMet substructure was solved using
the “half-baked” approach as implemented in the program SHELXD.29 Twenty-seven sites were found and
validated, based on the consistency with the Patterson
function and with the non-crystallographic symmetry,
by manual inspection of the crossword table provided
by SHELXD. This confirms that there are nine independent monomers in the asymmetric unit with a VM
value of 4.4 Å3/Da, corresponding to a high solvent content of ,70%. Diffraction data were manipulated with
the CCP4 suite of programs.25 The selenium sites were
refined and phases calculated using MLPHARE30 using
the inflection ðl2 Þ as the native data set. Solvent flattening, histogram matching and ninefold non-crystallographic symmetry (NCS) averaging were applied using
Structure of P. furiosus Galactokinase
Table 1. Phasing and refinement statistics
Resolution range (Å)
Space group
Unit cell parameters (Å)
No. of reflections (working/free)
Rcullis acentric (peak/remote)
Overall figure-of-merit
Protein non-H atoms
Galactose atoms
ADP atoms
Mg atoms
R-factora/Rfreeb
rms deviationsc
Bond lengths (Å)
Bond angles (deg.)
Average B (Å2)
10 –2.9
C2221
a ¼ 211:7, b ¼ 355:4,
c ¼ 165:5
127,499/6821
0.84/0.77
0.56
22,729
108
243
9
0.23/0.27
0.015
1.66
64
Ramachandran plot statisticsd
Most favoured regions (%)
89.9
Additional allowed regions (%)
10.1
Disallowed regions (%)
0
P
P
a
R-factor ¼ hkl llFo l 2 lFc ll= hkl lFo l:
b
Rfree is the crystallographic R value calculated using 5% of
the native data witheld from refinement.
c
Root-mean-square (rms) deviations are from ideal bond
lengths and angles, respectively.
d
Average over all subunits.
DM31 and an initial model was built into the MAD
phased map calculated to 2.9 Å using TURBO-FRODO.32
Iterative cycles of refinement and rebuilding using tight
NCS restraints until the final cycle, in REFMAC33 and
including TLS refinement, resulted in a final model of
22,729 atoms, with R ¼ 0:23 and Rfree ¼ 0:27 (Table 1).
The overall average B-factor is 64 Å2. In the final model
there are no non-glycine Ramachandran outliers, 90% of
the residues are in the most favoured conformation and
all 27 selenium sites correspond to the positions of
methionine residues. The Figures shown here have been
prepared using the following programs; Figure 1,
ALSCRIPT;34 Figures 2 and 4 – 6, MOLSCRIPT35 and
Raster3D;36 Figure 3, BOBSCRIPT37 and Raster3D36.
Accession numbers
The atomic coordinates for Pf GalK in complex with
galactose and ADP have been deposited in the RCSB
Protein Data Bank with accession number 1S4E.
Acknowledgements
We thank the support staff and Dr H. Belrhali of
beamline ID14.4 at the ESRF Grenoble laboratory
for assistance with station alignment. This work
was supported by the BBSRC, the Wellcome Trust
and the Leverhulme Trust. The Krebs Institute is a
designated BBSRC Biomolecular Science Centre
and a member of the North of England Structural
Biology Centre. A.H. is a BBSRC CASE student
397
Structure of P. furiosus Galactokinase
with British Biotechnology. C.V. was supported by
the Earth and Life Science Foundation (ALW),
which is subsidized by the Netherlands Organization for Scientific Research (NWO).
References
1. Frey, P. A. (1996). The Leloir pathway: a mechanistic
imperative for three enzymes to change the stereochemical configuration of a single carbon in
galactose. FASEB J. 10, 461– 470.
2. Petry, K. G. & Reichardt, J. K. V. (1998). The fundamental importance of human galactose metabolism:
lessons from genetics and biochemistry. Trends
Genet. 14, 98 – 102.
3. Stambolian, D., Scarpino-Myers, V., Hodes, B. &
Harris, H. (1986). Cataracts in patients heterozygous
for galactokinase deficiency. Invest. Opthal. Vis. Sci.
27, 429– 433.
4. Stambolian, D., Ai, Y., Sidjanin, D., Nesburn, K.,
Sathe, G., Rosenberg, M. & Bergsma, D. J. (1995).
Cloning of the galactokinase cDNA and identification of mutations in two families with cataracts.
Nature Genet. 10, 307– 312.
5. Kalaydjieva, L., Perez-Lezaun, A., Angelicheva, D.,
Onengut, S., Dye, D., Bosshard, N. U. et al. (1999). A
founder mutation in the GK1 gene is responsible for
galactokinase deficiency in Roma (Gypsies). Am.
J. Hum. Genet. 65, 1299– 1307.
6. Kolosha, V., Anoia, E., de Cespedes, C., Gitzelmann,
R., Shih, L., Casco, T., Saborio, M. et al. (2000). Novel
mutations in 13 probands with galactokinase
deficiency. Hum. Mutat. 15, 447– 453.
7. Hunter, M., Angelicheva, D., Levy, H. L., Peuschel,
S. M. & Kalaydjieva, L. (2001). Novel mutations in
the GALK1 gene in patients with galactokinase
deficiency. Hum. Mutat. 17, 77 – 78.
8. Okano, Y., Asada, M., Fujimoto, A., Ohtake, A.,
Murayama, K., Hsiao, K. J. et al. (2001). A genetic
factor for age-related cataract: identification and
characterization of a novel galactokinase variant,
“Osaka”, in Asians. Am. J. Hum. Genet. 68,
1036–1042.
9. Timson, D. & Reece, R. J. (2003). Functional analysis
of disease-causing mutations in human galactokinase. Eur. J. Biochem. 270, 1767–1774.
10. Johnston, M. (1987). A model fungal gene regulatory
mechanism: the GAL genes of Saccharomyces
cerevisiae. Microbiol. Rev. 51, 458– 476.
11. Reece, R. J. & Platt, A. (1997). Signaling activation
and repression of RNA polymerase II transcription
in yeast. Bioessays, 19, 1001– 1010.
12. Yano, K. & Fukasawa, T. (1996). Galactose-dependent
reversible interaction of Gal3p with Gal80p in the
induction pathway of Gal4p-activated genes of
Saccharomyces cerevisiae. Proc. Natl Acad. Sci. USA, 94,
1721–1726.
13. Platt, A., Ross, H. C., Hankin, S. & Reece, R. J. (2000).
The insertion of two amino acids into a transcriptional inducer converts it into a galactokinase. Proc.
Natl Acad. Sci. USA, 97, 3154– 3159.
14. Bork, P., Sander, C. & Valencia, A. (1993). Convergent
evolution of similar enzymatic function on different
protein folds—the hexokinase, ribokinase and
galactokinase families of sugar kinases. Protein Sci.
2, 31 –40.
15. Zhou, T., Daugherty, M., Grishin, N. V., Osterman,
A. L. & Zhang, H. (2000). Structure and mechanism
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
of homoserine kinase: prototype for the GHMP
kinase superfamily. Struct. Fold. Des. 8, 1247– 1257.
Fu, Z., Wang, M., Potter, D., Miziorko, H. M. & Kim,
J. (2002). The structure of a binary complex between
a mammalian mevalonate kinase and ATP: insights
into the reaction mechanism and inherited disease.
J. Biol. Chem. 277, 18134– 18142.
Yang, D., Shipma, L. W., Roessner, C. A., Scott, A. I.
& Sacchettini, J. C. (2002). Structure of the Methanococcus jannaschii mevalonate kinase, a member of the
GHMP kinase superfamily. J. Biol. Chem. 277,
9462 –9467.
Bonanno, J. B., Edo, C., Eswar, N., Pieper, U.,
Romanowski, M. J., Ilyin, V. et al. (2001). Structural
genomics of enzymes involved in sterol/isoprenoid
biosynthesis. Proc. Natl Acad. Sci. USA, 98,
12896 – 12901.
Romanowski, M. J., Bonanno, J. B. & Burley, S. K.
(2002). Crystal structure of the Streptococcus
pneumoniae phosphomevalonate kinase, a member of
the GHMP kinase superfamily. Proteins: Struct.
Funct. Genet. 47, 568– 571.
Krishna, S. S., Zhou, T., Daugherty, M., Osterman, A.
& Zhang, H. (2001). Structural basis for the catalysis
and substrate specificity of homoserine kinase.
Biochemistry, 40, 10810– 10818.
Thoden, J. B. & Holden, H. M. (2003). Molecular
structure of galactokinase. J. Biol. Chem. 278,
33305 – 33311.
Jones, S. & Thornton, J. M. (1995). Protein – protein
interactions: a review of protein dimer structures.
Prog. Biophys. Mol. Biol. 63, 31 –65.
De Geus, D., Hartley, A. P., Sedelnikova, S. E., Glynn,
S. E., Baker, P. J., Verhees, C. et al. (2003). Cloning,
purification, crystallization and preliminary crystallographic analysis of galactokinase from Pyrococcus
furiosus. Acta Crystallog. sect. D, 59, 1819– 1821.
Verhees, C. H., Koot, D. G., Ettema, T. J., Dijkema, C.,
de Vos, W. M. & van der Oost, J. (2002). Biochemical
adaptations of two sugar kinases from the hyperthermophilic archaeon Pyrococcus furiosus. Biochem. J.
366, 121– 127.
Collaborative Computational Project No. 4 (1994).
The CCP4 suite: programs for protein crystallography. Acta Crystallog. sect. D, 50, 760– 763.
Zenke, F. T., Engles, R., Vollenbroich, V., Meyer, J.,
Hollenberg, C. P. & Breunig, K. D. (1996). Activation
of Gal4p by galactose-dependent interaction of
galactokinase and Gal80p. Science, 272, 1662– 1665.
Timson, D. J., Ross, H. C. & Reece, R. J. (2002). Gal3p
and Gal1p interact with the transcriptional repressor
Gal80p to form a complex of 1:1 stoichiometry.
Biochem. J. 363, 515–520.
Otwinowski, Z. & Minor, W. (1997). Processing of
X-ray diffraction data collected in oscillation mode.
Methods Enzymol. 276, 307– 326.
Sheldrick, G. M. (1998). SHELX: applications to
macromolecules. In Direct Methods for Solving Macromolecular Structures (Fortier, S., ed.), pp. 401– 411,
Kluwer Academic, Dordrecht.
Otwinowski, Z. (1991). Maximum likelihood refinement of heavy atom parameters in isomorphous
replacement and anomalous scattering. In Proceeding
of the CCP4 Study Weekend (Wolf, W., Evans, P. R. &
Leslie, A. G. W., eds), pp. 80 – 86, SERC Daresbury
Laboratory, Warrington, UK.
Cowtan, K. (1994). DM: an automated procedure for
phase improvement by density modification. Joint
398
Structure of P. furiosus Galactokinase
CCP4 ESF-EACBM Newsletter Protein Crystallog. 31,
34 – 38.
32. Roussel, A. & Cambillau, C. (1991). Silicon Graphics
Geometry Partners Directory 86, Silicon Graphics,
Mountain View, CA.
33. Murshodov, G. N., Vagin, A. A. & Dodson, E. J.
(1997). Refinement of macromolecular structures by
the maximum likelihood method. Acta Crystallog.
sect. D, 53, 240– 255.
34. Barton, G. J. (1993). ALSCRIPT: a tool to format
multiple sequence alignments. Proteins Eng. 6, 37 – 40.
35. Kraulis, P. J. (1991). MOLSCRIPT: a program to
produce both detailed and schematic plots of protein
structures. J. Appl. Crystallog. 24, 946– 950.
36. Merrit, E. A. & Bacon, D. J. (1994). Raster3D: photorealistic molecular graphics. Methods Enzymol. 277,
505– 524.
37. Esnouf, R. M. (1999). Further additions to MolScript
version 1.4, including reading and contouring of
electron-density maps. Acta Crystallog. sect. D, 55,
938– 940.
Edited by D. Rees
(Received 15 August 2003; received in revised form 16 January 2004; accepted 24 January 2004)